T Pipeline Demos

This page lists real-world T projects that demonstrate the power of polyglot, reproducible orchestration. Most of these demos are available in the tstats-project/t_demos repository.

Visualization & Reporting

Plotting & Visual Metadata Capture

Repo: plotting_pipeline_t
Description: Demonstrates how T captures structural metadata from ggplot2 (R) and matplotlib (Python).
Key Features: Polyglot plots, metadata extraction, Nix-reproducible charts.

Quarto Literate Programming

Repo: quarto_test_t
Description: Shows how to embed T pipeline results in a Quarto report.
Key Features: Auto-substitution of read_node(), Nix-built reports.

Statistical Modeling

Model Comparison: R vs Python

Repo: model_comparison_t
Description: Compares GLM results across R and Python nodes within the same pipeline.
Key Features: Concurrent R and Python environments, PMML interchange.

Titanic Survival

Repo: glm_titanic_t
Description: A classic statistical modeling pipeline using the Titanic dataset.
Key Features: Data cleaning, logistic regression, result serialization.

Advanced Orchestration

ONNX & PMML Interchange

Repo: onnx_exchange_t / pmml_interchange_t
Description: Moving trained models between high-level languages and T’s native scoring engine.
Key Features: predict() across language boundaries, standardized model storage.

Lenses & Dynamic Pipelines

Repo: deep_data_lenses_t / dynamic_pipeline_operator_t
Description: Demonstrates functional updates to immutable configurations and dynamic DAG generation.
Key Features: Pipeline-as-code, functional composition.

Meta-programming & Introspection

Dynamic Lookup & Symbols

Repo: get_sym_demo_t
Description: Showcases runtime variable lookup and symbol construction for dynamic orchestration.
Key Features: get(), sym(), dynamic node access.

Running the Demos

To try any of these demos locally:

Clone the demos repository:

git clone https://github.com/tstats-project/t_demos
cd t_demos/<project_name>

Bootstrap the T environment: If you don’t have the t command installed yet, use Nix to get a temporary shell containing it:
```
nix shell github:b-rodrigues/tlang
```
Synchronize dependencies: T uses tproject.toml to manage dependencies. Run this to generate the project’s local flake.nix:
```
t update
```
Once t update is complete, exit the temporary shell:
```
exit
```
Enter the project shell: Now use the project-specific environment generated by T:
```
nix develop
```
Run the pipeline:
```
t run src/pipeline.t
```